Full text of DOI:10.1557/JMR.2002.0361

Epitaxial growth of ZnO films on Si(111)
Ashutosh Tiwari,a) M. Park, C. Jin, H. Wang, D. Kumar, and J. Narayan
Department of Materials Science and Engineering, North Carolina State University,
Raleigh, North Carolina 27695-7916
(Received 20 May 2002; accepted 23 July 2002)
In this paper, we report the growth of ZnO films on silicon substrates using a pulsed
laser deposition technique. These films were deposited on Si(111) directly as well as
by using thin buffer layers of AlN and GaN. All the films were found to have
c-axis-preferred orientation aligned with normal to the substrate. Films with AlN and
GaN buffer layers were epitaxial with preferred in-plane orientation, while those
directly grown on Si(111) were found to have random in-plane orientation. A decrease
(2)
in the frequency of the E2 Raman mode and a red shift of the band-edge
photoluminescence peak due to the presence of tensile strain in the film, was observed.
Various possible sources for the observed biaxial strain are discussed.
During the last few years wide-band-gap semiconduc-
tors have attracted significant scientific and technologi-
cal interest.1–3 ZnO holds a special position in materials
physics due to its interesting electrical and optical char-
acteristics.1,2 Because of its higher exciton binding en-
ergy (60 meV) and higher luminescent efficiencies it is
considered a substitute for GaN of the III–V semicon-
ductor family3 for various optical and light emitting ap-
plications. Its band gap (3.28 eV) is quite close to that of
GaN (3.44 eV). The band gap of GaN can be increased or
decreased by alloying with AlN (6.2 eV) or with InN
(1.9 eV). Similarly, ZnO can be alloyed with MgO
(8.2 eV) to increase the band gap or with CdO (2.0 eV) to
decrease it.
For the use of ZnO in light emitting and nonlinear
optical applications, high-quality single-crystal ZnO
films are required. Epitaxial films of ZnO have been
previously grown on (0001) planes of sapphire by the
domain epitaxy mechanism.4,5 The growth of high-
quality ZnO films on silicon is expected to offer addi-
tional advantages as it will provide opportunity to
integrate various functional properties of ZnO with the
advanced silicon-based microelectronics devices. How-
ever, direct growth of epitaxial ZnO films on silicon is
extremely difficult; often it results in inferior quality
films.6 Here, we report the epitaxial growth of ZnO films
by employing buffer layers of AlN and GaN in a pulsed
laser deposition process. This is the first report of growth
of epitaxial ZnO on Si(111) with an AlN buffer layer;
however, epitaxial growth of ZnO on Si(111)/GaN has
been reported earlier.7 We have also deposited ZnO di-
rectly on hydrogen-terminated Si(111) surfaces.
One of the most crucial factors that affect the perfor-
mance of thin-film devices is the presence of biaxial
strain (and associated stress) in the films.8,9 It has been
realized that the band structure of ZnO may change with
the strain and subsequently may modify the optical and
electrical characteristics of the film. We have investi-
gated the strain present in these films as reflected on
photoluminescence and Raman scattering measurements.
The shift in the value of photoluminescence band-edge
energy and the position of the E2(2) Raman active phonon
mode is used to characterize the sign and the relative
magnitude of the stress present in these films.
Thin-film deposition was performed in a multitarget
laser deposition chamber with a KrF excimer laser
(Lambda Physik 210, ␭ ס 248 nm). Energy density
and repetition rate of the laser beam were 2–3 J/cm2 and
10 Hz, respectively. Before deposition, Si(111) sub-
strates were treated with 5% hydrofluoric acid solution to
remove the surface oxide layer. AlN and GaN films were
deposited in the vacuum (1 × 10−7 torr) at 700 °C, while
ZnO films were deposited at 650 °C in an ambient oxy-
gen pressure of 1 × 10−5 torr. The thickness of the AlN
and GaN layers was approximately 100 nm, and that of
the ZnO layer was approximately 500 nm. The thickness of
individual layers was precalibrated against the number
of pulses. X-ray diffraction (XRD) measurements were
performed using Cu K␣ radiation and a nickel filter. Se-
lected area electron diffraction patterns were obtained by
cross-sectional transmission electron microscopy (TEM)
using a JEOL-2010F analytical microscope. Micro Ra-
man spectroscopy measurements were carried out at
a)Address all correspondence to this author.
e-mail: atiwari@unity.ncsu.edu
2480
J. Mater. Res., Vol. 17, No. 10, Oct 2002
© 2002 Materials Research Society
http://journals.cambridge.org
Downloaded: 16 Feb 2015
IP address: 130.179.16.201

Rapid Communications
room temperature in the backscattering geometry with
the 514.5-nm (2.41 eV) line of an Ar ion laser using an
ISA U-1000 scanning double monochromator (JOBIN
YVON, USA). Laser power of 120–140 mW was used to
excite the sample. It was ensured that the excitation laser
beam did not cause any heating of the sample. Photolu-
minescence (PL) measurements were performed using a
Hitachi F-4500 FL spectrophotometer.
Figure 1 shows the XRD pattern of ZnO films grown
on Si(111). In all the three cases we could obtain only the
peaks corresponding to the (0002) planes of ZnO, AlN,
GaN, the (111) peak of the Si, and the (0004) peak of
ZnO. This indicates that all the films have c-axis-
preferred orientation. However, the XRD reveals the
alignment of the films only in the direction normal to
the film. To determine the in-plane alignment of these
films, we performed TEM investigations on various parts
of these films. We found that while the films grown on
Si(111) with AlN and GaN buffer layers are epitaxial,
those grown directly on Si(111) show random in-plane
orientation. In Fig. 2 we show the selected area diffrac-
tion pattern from ZnO film in the Si(111)/AlN/ZnO het-
erostructure. Well-aligned spots clearly establish the
epitaxial nature of ZnO films. Figure 2(b) shows the dif-
fraction pattern from the ZnO/AlN interface. Comparing
this with standard diffraction patterns we found that ZnO
and AlN have the following orientational relationship:
¯ ¯
¯ ¯
ZnO[2110]࿣AlN[2110] along the in-plane direction
and ZnO[0001]࿣AlN[0001] along the normal to the
plane. ZnO and GaN layers in the Si(111)/GaN/ZnO het-
erostructure were found to have similar orienta-
tional relationships.
Figure 3(a) shows the Raman spectra of ZnO films
deposited on various substrates. Raman spectra of
Si(111) substrate and single-crystal ZnO are also shown
FIG. 2. Selected area diffraction pattern from cross section TEM
sample of ZnO/AlN/Si (a) ZnO film only (b) ZnO/AlN interface.
for comparison. Since the ZnO, GaN, and AlN films are
transparent in the 514-nm excitation, the intense Raman
peak (520 cm−1) from Si substrate was observed in all the
three samples. The Raman spectra indicate that the ZnO
films have a wurtzite structure. The wurtzite form of
ZnO belongs to the C6␯4 (P63mc) space group. According
to the group theoretical analysis, the following optical
modes exist at the ⌫ point of the Brillouin zone of the ZnO:
FIG. 1. X-ray diffraction patterns of Si(111)/AlN/ZnO, Si(111)/GaN/
ZnO, and Si(111)/ZnO.
⌫ ס A1(TO,LO) + 2B1 + E1(TO,LO) + 2E2
.
J. Mater. Res., Vol. 17, No. 10, Oct 2002
2481
http://journals.cambridge.org
Downloaded: 16 Feb 2015
IP address: 130.179.16.201

Rapid Communications
Our analysis showed that the peak at around 437 cm−1
(2)
corresponds to the E2 mode of ZnO. We fitted it with
a Lorentzian function using a linear background. The
position and full width at half-maximum (FWHM) of the
(2)
Raman E2 peaks are shown in Fig. 3(b). As is evident
(2)
from this figure, the E2 peak shifts to the lower fre-
quency side as the substrate is changed from AlN/Si to Si
and GaN/Si. The peak position obtained from the single-
crystal ZnO was used as the reference point. A decrease
(2)
in the frequency shift of the E2 mode suggests the
presence of tensile stress (and associated strain) in these
films. By comparing our results with those of Mitra
et al.,11 we made an estimate that few kbars of tensile
(2)
stress exist in our films. The FWHM of E2 peak de-
pends on the quality of the sample. As can be seen in
Fig. 3(b), the FWHM of E2(2) peaks for our films is about
30–50% higher than that for ZnO single crystal. This
indicates that the crystal quality of bulk single crystal is
much higher than that of films.
In Fig. 4 we show the room-temperature PL spectra
of the ZnO films. For comparison, we have also shown
the room-temperature PL spectrum of single-crystal
ZnO. Because of the large exciton binding energy,
60 meV, excitonic emission survives even above room
temperature. The band-edge maximum PL intensity peak
for self-standing stress-free single-crystal ZnO films and
bulk has been reported12 to occur at about 3.285 eV. The
PL peaks of the films in our study occur at slightly lower
energies. This red shift in the PL peak is in agreement
with the tensile strain as observed in Raman spectrum. At
lower energies we observed a very broad PL peak. This
broad peak has been observed previously also by several
workers and is usually referred as the green band.13 Rey-
nolds et al.13 attributed the origin of the green band to
FIG. 3. (a) Raman spectra of various ZnO films. Raman spectra of
Si(111) substrate and single-crystal ZnO are also shown for compari-
(2)
son. (b) Position and FWHM of Raman E2 peaks for various ZnO
samples.
Among these optic modes, two B1 modes are silent, and
A1, E1, and E2 modes are Raman active. The A1 and E1
modes further split into transverse and longitudinal com-
ponent due to their polar nature. The A1 and E1 modes
are polarized along and perpendicular to the c axis of the
hexagonal unit cell of ZnO, respectively. The E2 modes
have a low- and high-frequency submode, which are
(1)
(2)
commonly designated as E2 and E2 modes, respec-
(2)
tively. It has been observed that the position of the E2
FIG. 4. Room-temperature PL spectrum of various ZnO films. For
comparison, we have also shown the room-temperature PL spectrum
of single-crystal ZnO.
peaks is sensitive to any kind of pressure or stress in
the film.10
2482
J. Mater. Res., Vol. 17, No. 10, Oct 2002
http://journals.cambridge.org
Downloaded: 16 Feb 2015
IP address: 130.179.16.201

Rapid Communications
defect states in the material. These defect states could
interstitials, dislocations, etc. If vacancies with lower
specific volume compared to the host lattice are present,
then a net tensile stress prevails in the film. Oxygen
vacancies in ZnO films are expected to cause tensile
strain. Since the net strain in the ZnO films, as deter-
be due to the presence of oxygen vacancies in the
film. The intensity of the green band is greatest for
the Si(111)/ZnO; this shows that the films directly grown
on Si(111) have a greater number of defects and/or oxy-
gen vacancies.
(2)
mined by a decrease in the frequency of E2 mode of
Strain in thin-film systems usually has the following
three origins: (i) the lattice mismatch between the film
and substrate (or buffer layer); (ii) difference in the ther-
mal expansion coefficient of the film and substrate (or
buffer layer); (iii) microstructure/defect related internal
stresses. The lattice mismatch arises due to the differ-
ences in lattice constants between the film and the sub-
strate under epitaxial growth conditions. If the lattice
constant of the film af is less than the lattice parameter of
substrate (af < as), then there is a tensile stress in the
film; on the other hand, if (af > as), then the film is under
compressive strain. Under the lattice matching, total
strain is given by ⑀T ס 2(af − as). Since the in-plane lat-
tice parameter of ZnO is more than that of AlN and GaN
(see Table I), lattice mismatch strain in ZnO is expected
to be compressive in the case of Si(111)/AlN/ZnO and
Si(111)/GaN/ZnO heterostructures. The film directly
grown on Si(111) has random in-plane orientation, and
hence lattice mismatch strain is expected to be absent in
this film. Furthermore, it has been reported that beyond a
critical thickness of the film most of the lattice mismatch
stress gets dissipated in creating dislocations in the film
at the growth temperature. However, as the film cools
down, the thermal strain is developed in the film due to
the difference in the thermal expansion coefficient of the
substrate and the film. If the coefficient of thermal ex-
pansion of the films ␣f is greater than the coefficient
of thermal expansion of the substrate ␣s (␣f > ␣s), then
the thermal strain in the film is expected to be tensile;
on the other hand, if ␣f < ␣s, then the thermal strain will
be compressive in nature. As can be seen in Table I, the
value of the coefficient of thermal expansion of ZnO
along the a–b basal plane is less than those of AlN, GaN,
and Si; this implies that the thermal strain in ZnO film
should be compressive. The third cause of the strain,
namely, the microstructure/defect related internal stress,
is resulting from the trapped point defects like vacancies,
Raman spectra and a red-shift of the band-edge PL peak,
is tensile, we predict that the microstructure/defect re-
lated internal stress is the main source of the strain in
the system.
In conclusion, we have grown and integrated single-
crystal epitaxial ZnO films on Si(111) substrates by em-
ploying AlN and GaN buffer layers in a pulsed laser
deposition process. A thorough investigation of their op-
tical characteristics has been carried out using Raman
scattering and PL experiments. Growth of epitaxial ZnO
films on silicon substrates with a precise understanding
of their optical characteristics provides excellent oppor-
tunity to integrate various functional properties of ZnO
with the silicon-based microelectronics devices.
ACKNOWLEDGMENTS
This work was partially supported by the National Sci-
ence Foundation. The authors thank Prof. R.J. Nemanich
for very useful comments and permission to use the
micro-Raman facilities.
REFERENCES
1. H. Ohno, Science 281, 951, 1998.
2. Z.K. Tang, P. Yu, G.K.L. Wong, M. Kawasaki, A. Ohtomo,
H. Koinuma, and Y. Segawa, Solid State Commun. 103, 459
(1997).
3. H. Morkoc¸, S. Strite, G.B. Gao, M.E. Lin, B. Sverdloy, and
M. Burns, J. Appl. Phys. 76, 1363 (1994).
4. A. Tiwari, C. Jin, A. Kvit, D. Kumar, J.F. Muth, and J. Narayan,
Solid State Commun. 121, 371 (2002).
5. G.J. Exarhos and S.K. Sharma, Thin Solid Films 270, 27 (1995).
6. T. Shirasawa, T. Honda, F. Koyama, and K. Iga, in III-V Nitrides,
edited by F.A. Ponce, T.D. Moustakas, J. Akasaki, and
B.A. Monemar (Mater. Res. Soc. Symp. Proc. 449, Warrendale,
PA, 1997), p. 373.
7. A. Nahhas, H.K. Kim, and J. Blachere, Appl. Phys. Lett. 78, 1511,
2001.
8. B.H. Bairamov, O. Gu¨rdal, A. Botchkarev, H. Morkoc¸, G. Irmer,
and J. Monecke, Phys. Rev. B 60, 16741 (1999).
9. A. Konkar, A. Madhukar, and P. Chen, Appl. Phys. Lett. 72, 220
(1998).
TABLE I. Lattice parameters and the coefficients of thermal expan-
sion of Si, AlN, GaN, and ZnO along the a–b basal plane.
10. F. Decremps, J. Pellicer-Porres, A.M. Saitta, J-C. Chervin, and
A. Polian, Phys. Rev. B 65, 92101 (2002).
11. S.S. Mitra, O. Brafman, W.B. Daniels, and R.K. Crawford, Phys.
Thermal expansion coefficient
Lattice parameter (Å)
(K−1
)
3.6 × 10−6
4.5 × 10−6
5.6 × 10−6
2.9 × 10−6
Rev. B 186, 942 (1969).
12. F. Decremps, J. Pellicer-Porres, A.M. Saitta, J-C. Chervin, and
A. Polian, Phys. Rev. B 65, 92101 (2002).
Si
a ס 5.431
AlN
GaN
ZnO
a ס 3.112; c ס 5.186
a ס 3.189; c ס 5.185
a ס 3.252; c ס 5.213
13. D.C. Reynolds, D.C. Look, B. Jogai, and H. Morkoc¸, Solid State
Commun. 101(9), 643 (1997).
J. Mater. Res., Vol. 17, No. 10, Oct 2002
2483
http://journals.cambridge.org
Downloaded: 16 Feb 2015
IP address: 130.179.16.201